Bison is a general-purpose parser generator that converts a grammar description for an LALR(1) context-free grammar into a C program to parse that grammar. Once you are proficient with Bison, you may use it to develop a wide range of language parsers, from those used in simple desk calculators to complex programming languages.
Bison is upward compatible with Yacc: all properly-written Yacc grammars ought to work with Bison with no change. Anyone familiar with Yacc should be able to use Bison with little trouble. You need to be fluent in C programming in order to use Bison or to understand this manual.
We begin with tutorial chapters that explain the basic concepts of using Bison and show three explained examples, each building on the last. If you don't know Bison or Yacc, start by reading these chapters. Reference chapters follow which describe specific aspects of Bison in detail.
Bison was written primarily by Robert Corbett; Richard Stallman made it Yacc-compatible. Wilfred Hansen of Carnegie Mellon University added multicharacter string literals and other features.
This edition corresponds to version 1.25 of Bison.
As of Bison version 1.24, we have changed the distribution terms for
yyparse to permit using Bison's output in non-free programs.
Formerly, Bison parsers could be used only in programs that were free
software.
The other GNU programming tools, such as the GNU C compiler, have never had such a requirement. They could always be used for non-free software. The reason Bison was different was not due to a special policy decision; it resulted from applying the usual General Public License to all of the Bison source code.
The output of the Bison utility--the Bison parser file--contains a
verbatim copy of a sizable piece of Bison, which is the code for the
yyparse function. (The actions from your grammar are inserted
into this function at one point, but the rest of the function is not
changed.) When we applied the GPL terms to the code for yyparse,
the effect was to restrict the use of Bison output to free software.
We didn't change the terms because of sympathy for people who want to make software proprietary. Software should be free. But we concluded that limiting Bison's use to free software was doing little to encourage people to make other software free. So we decided to make the practical conditions for using Bison match the practical conditions for using the other GNU tools.
Version 2, June 1991
Copyright (C) 1989, 1991 Free Software Foundation, Inc. 675 Mass Ave, Cambridge, MA 02139, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.
The precise terms and conditions for copying, distribution and modification follow.
NO WARRANTY
If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found.
one line to give the program's name and a brief idea of what it does. Copyright (C) 19yy name of author This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details.
The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program.
You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. signature of Ty Coon, 1 April 1989 Ty Coon, President of Vice
This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License.
This chapter introduces many of the basic concepts without which the details of Bison will not make sense. If you do not already know how to use Bison or Yacc, we suggest you start by reading this chapter carefully.
In order for Bison to parse a language, it must be described by a context-free grammar. This means that you specify one or more syntactic groupings and give rules for constructing them from their parts. For example, in the C language, one kind of grouping is called an `expression'. One rule for making an expression might be, "An expression can be made of a minus sign and another expression". Another would be, "An expression can be an integer". As you can see, rules are often recursive, but there must be at least one rule which leads out of the recursion.
The most common formal system for presenting such rules for humans to read is Backus-Naur Form or "BNF", which was developed in order to specify the language Algol 60. Any grammar expressed in BNF is a context-free grammar. The input to Bison is essentially machine-readable BNF.
Not all context-free languages can be handled by Bison, only those that are LALR(1). In brief, this means that it must be possible to tell how to parse any portion of an input string with just a single token of look-ahead. Strictly speaking, that is a description of an LR(1) grammar, and LALR(1) involves additional restrictions that are hard to explain simply; but it is rare in actual practice to find an LR(1) grammar that fails to be LALR(1). See section Mysterious Reduce/Reduce Conflicts, for more information on this.
In the formal grammatical rules for a language, each kind of syntactic unit or grouping is named by a symbol. Those which are built by grouping smaller constructs according to grammatical rules are called nonterminal symbols; those which can't be subdivided are called terminal symbols or token types. We call a piece of input corresponding to a single terminal symbol a token, and a piece corresponding to a single nonterminal symbol a grouping.
We can use the C language as an example of what symbols, terminal and nonterminal, mean. The tokens of C are identifiers, constants (numeric and string), and the various keywords, arithmetic operators and punctuation marks. So the terminal symbols of a grammar for C include `identifier', `number', `string', plus one symbol for each keyword, operator or punctuation mark: `if', `return', `const', `static', `int', `char', `plus-sign', `open-brace', `close-brace', `comma' and many more. (These tokens can be subdivided into characters, but that is a matter of lexicography, not grammar.)
Here is a simple C function subdivided into tokens:
int /* keyword `int' */
square (x) /* identifier, open-paren, */
/* identifier, close-paren */
int x; /* keyword `int', identifier, semicolon */
{ /* open-brace */
return x * x; /* keyword `return', identifier, */
/* asterisk, identifier, semicolon */
} /* close-brace */
The syntactic groupings of C include the expression, the statement, the declaration, and the function definition. These are represented in the grammar of C by nonterminal symbols `expression', `statement', `declaration' and `function definition'. The full grammar uses dozens of additional language constructs, each with its own nonterminal symbol, in order to express the meanings of these four. The example above is a function definition; it contains one declaration, and one statement. In the statement, each `x' is an expression and so is `x * x'.
Each nonterminal symbol must have grammatical rules showing how it is made
out of simpler constructs. For example, one kind of C statement is the
return statement; this would be described with a grammar rule which
reads informally as follows:
A `statement' can be made of a `return' keyword, an `expression' and a `semicolon'.
There would be many other rules for `statement', one for each kind of statement in C.
One nonterminal symbol must be distinguished as the special one which defines a complete utterance in the language. It is called the start symbol. In a compiler, this means a complete input program. In the C language, the nonterminal symbol `sequence of definitions and declarations' plays this role.
For example, `1 + 2' is a valid C expression--a valid part of a C program--but it is not valid as an entire C program. In the context-free grammar of C, this follows from the fact that `expression' is not the start symbol.
The Bison parser reads a sequence of tokens as its input, and groups the tokens using the grammar rules. If the input is valid, the end result is that the entire token sequence reduces to a single grouping whose symbol is the grammar's start symbol. If we use a grammar for C, the entire input must be a `sequence of definitions and declarations'. If not, the parser reports a syntax error.
A formal grammar is a mathematical construct. To define the language for Bison, you must write a file expressing the grammar in Bison syntax: a Bison grammar file. See section Bison Grammar Files.
A nonterminal symbol in the formal grammar is represented in Bison input
as an identifier, like an identifier in C. By convention, it should be
in lower case, such as expr, stmt or declaration.
The Bison representation for a terminal symbol is also called a token
type. Token types as well can be represented as C-like identifiers. By
convention, these identifiers should be upper case to distinguish them from
nonterminals: for example, INTEGER, IDENTIFIER, IF or
RETURN. A terminal symbol that stands for a particular keyword in
the language should be named after that keyword converted to upper case.
The terminal symbol error is reserved for error recovery.
See section Symbols, Terminal and Nonterminal.
A terminal symbol can also be represented as a character literal, just like a C character constant. You should do this whenever a token is just a single character (parenthesis, plus-sign, etc.): use that same character in a literal as the terminal symbol for that token.
A third way to represent a terminal symbol is with a C string constant containing several characters. See section Symbols, Terminal and Nonterminal, for more information.
The grammar rules also have an expression in Bison syntax. For example,
here is the Bison rule for a C return statement. The semicolon in
quotes is a literal character token, representing part of the C syntax for
the statement; the naked semicolon, and the colon, are Bison punctuation
used in every rule.
stmt: RETURN expr ';'
;
See section Syntax of Grammar Rules.
A formal grammar selects tokens only by their classifications: for example, if a rule mentions the terminal symbol `integer constant', it means that any integer constant is grammatically valid in that position. The precise value of the constant is irrelevant to how to parse the input: if `x+4' is grammatical then `x+1' or `x+3989' is equally grammatical.
But the precise value is very important for what the input means once it is parsed. A compiler is useless if it fails to distinguish between 4, 1 and 3989 as constants in the program! Therefore, each token in a Bison grammar has both a token type and a semantic value. See section Defining Language Semantics, for details.
The token type is a terminal symbol defined in the grammar, such as
INTEGER, IDENTIFIER or ','. It tells everything
you need to know to decide where the token may validly appear and how to
group it with other tokens. The grammar rules know nothing about tokens
except their types.
The semantic value has all the rest of the information about the
meaning of the token, such as the value of an integer, or the name of an
identifier. (A token such as ',' which is just punctuation doesn't
need to have any semantic value.)
For example, an input token might be classified as token type
INTEGER and have the semantic value 4. Another input token might
have the same token type INTEGER but value 3989. When a grammar
rule says that INTEGER is allowed, either of these tokens is
acceptable because each is an INTEGER. When the parser accepts the
token, it keeps track of the token's semantic value.
Each grouping can also have a semantic value as well as its nonterminal symbol. For example, in a calculator, an expression typically has a semantic value that is a number. In a compiler for a programming language, an expression typically has a semantic value that is a tree structure describing the meaning of the expression.
In order to be useful, a program must do more than parse input; it must also produce some output based on the input. In a Bison grammar, a grammar rule can have an action made up of C statements. Each time the parser recognizes a match for that rule, the action is executed. See section Actions. Most of the time, the purpose of an action is to compute the semantic value of the whole construct from the semantic values of its parts. For example, suppose we have a rule which says an expression can be the sum of two expressions. When the parser recognizes such a sum, each of the subexpressions has a semantic value which describes how it was built up. The action for this rule should create a similar sort of value for the newly recognized larger expression.
For example, here is a rule that says an expression can be the sum of two subexpressions:
expr: expr '+' expr { $$ = $1 + $3; }
;
The action says how to produce the semantic value of the sum expression from the values of the two subexpressions.
When you run Bison, you give it a Bison grammar file as input. The output is a C source file that parses the language described by the grammar. This file is called a Bison parser. Keep in mind that the Bison utility and the Bison parser are two distinct programs: the Bison utility is a program whose output is the Bison parser that becomes part of your program.
The job of the Bison parser is to group tokens into groupings according to the grammar rules--for example, to build identifiers and operators into expressions. As it does this, it runs the actions for the grammar rules it uses.
The tokens come from a function called the lexical analyzer that you
must supply in some fashion (such as by writing it in C). The Bison parser
calls the lexical analyzer each time it wants a new token. It doesn't know
what is "inside" the tokens (though their semantic values may reflect
this). Typically the lexical analyzer makes the tokens by parsing
characters of text, but Bison does not depend on this. See section The Lexical Analyzer Function yylex.
The Bison parser file is C code which defines a function named
yyparse which implements that grammar. This function does not make
a complete C program: you must supply some additional functions. One is
the lexical analyzer. Another is an error-reporting function which the
parser calls to report an error. In addition, a complete C program must
start with a function called main; you have to provide this, and
arrange for it to call yyparse or the parser will never run.
See section Parser C-Language Interface.
Aside from the token type names and the symbols in the actions you
write, all variable and function names used in the Bison parser file
begin with `yy' or `YY'. This includes interface functions
such as the lexical analyzer function yylex, the error reporting
function yyerror and the parser function yyparse itself.
This also includes numerous identifiers used for internal purposes.
Therefore, you should avoid using C identifiers starting with `yy'
or `YY' in the Bison grammar file except for the ones defined in
this manual.
The actual language-design process using Bison, from grammar specification to a working compiler or interpreter, has these parts:
yylex). It could also be produced using Lex, but the use
of Lex is not discussed in this manual.
To turn this source code as written into a runnable program, you must follow these steps:
The input file for the Bison utility is a Bison grammar file. The general form of a Bison grammar file is as follows:
%{
C declarations
%}
Bison declarations
%%
Grammar rules
%%
Additional C code
The `%%', `%{' and `%}' are punctuation that appears in every Bison grammar file to separate the sections.
The C declarations may define types and variables used in the actions.
You can also use preprocessor commands to define macros used there, and use
#include to include header files that do any of these things.
The Bison declarations declare the names of the terminal and nonterminal symbols, and may also describe operator precedence and the data types of semantic values of various symbols.
The grammar rules define how to construct each nonterminal symbol from its parts.
The additional C code can contain any C code you want to use. Often the
definition of the lexical analyzer yylex goes here, plus subroutines
called by the actions in the grammar rules. In a simple program, all the
rest of the program can go here.
Now we show and explain three sample programs written using Bison: a reverse polish notation calculator, an algebraic (infix) notation calculator, and a multi-function calculator. All three have been tested under BSD Unix 4.3; each produces a usable, though limited, interactive desk-top calculator.
These examples are simple, but Bison grammars for real programming languages are written the same way.
The first example is that of a simple double-precision reverse polish notation calculator (a calculator using postfix operators). This example provides a good starting point, since operator precedence is not an issue. The second example will illustrate how operator precedence is handled.
The source code for this calculator is named `rpcalc.y'. The `.y' extension is a convention used for Bison input files.
rpcalcHere are the C and Bison declarations for the reverse polish notation calculator. As in C, comments are placed between `/*...*/'.
/* Reverse polish notation calculator. */
%{
#define YYSTYPE double
#include <math.h>
%}
%token NUM
%% /* Grammar rules and actions follow */
The C declarations section (see section The C Declarations Section) contains two preprocessor directives.
The #define directive defines the macro YYSTYPE, thus
specifying the C data type for semantic values of both tokens and groupings
(see section Data Types of Semantic Values). The Bison parser will use whatever type
YYSTYPE is defined as; if you don't define it, int is the
default. Because we specify double, each token and each expression
has an associated value, which is a floating point number.
The #include directive is used to declare the exponentiation
function pow.
The second section, Bison declarations, provides information to Bison about
the token types (see section The Bison Declarations Section). Each terminal symbol that is
not a single-character literal must be declared here. (Single-character
literals normally don't need to be declared.) In this example, all the
arithmetic operators are designated by single-character literals, so the
only terminal symbol that needs to be declared is NUM, the token
type for numeric constants.
rpcalcHere are the grammar rules for the reverse polish notation calculator.
input: /* empty */
| input line
;
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
exp: NUM { $$ = $1; }
| exp exp '+' { $$ = $1 + $2; }
| exp exp '-' { $$ = $1 - $2; }
| exp exp '*' { $$ = $1 * $2; }
| exp exp '/' { $$ = $1 / $2; }
/* Exponentiation */
| exp exp '^' { $$ = pow ($1, $2); }
/* Unary minus */
| exp 'n' { $$ = -$1; }
;
%%
The groupings of the rpcalc "language" defined here are the expression
(given the name exp), the line of input (line), and the
complete input transcript (input). Each of these nonterminal
symbols has several alternate rules, joined by the `|' punctuator
which is read as "or". The following sections explain what these rules
mean.
The semantics of the language is determined by the actions taken when a grouping is recognized. The actions are the C code that appears inside braces. See section Actions.
You must specify these actions in C, but Bison provides the means for
passing semantic values between the rules. In each action, the
pseudo-variable $$ stands for the semantic value for the grouping
that the rule is going to construct. Assigning a value to $$ is the
main job of most actions. The semantic values of the components of the
rule are referred to as $1, $2, and so on.
input
Consider the definition of input:
input: /* empty */
| input line
;
This definition reads as follows: "A complete input is either an empty
string, or a complete input followed by an input line". Notice that
"complete input" is defined in terms of itself. This definition is said
to be left recursive since input appears always as the
leftmost symbol in the sequence. See section Recursive Rules.
The first alternative is empty because there are no symbols between the
colon and the first `|'; this means that input can match an
empty string of input (no tokens). We write the rules this way because it
is legitimate to type Ctrl-d right after you start the calculator.
It's conventional to put an empty alternative first and write the comment
`/* empty */' in it.
The second alternate rule (input line) handles all nontrivial input.
It means, "After reading any number of lines, read one more line if
possible." The left recursion makes this rule into a loop. Since the
first alternative matches empty input, the loop can be executed zero or
more times.
The parser function yyparse continues to process input until a
grammatical error is seen or the lexical analyzer says there are no more
input tokens; we will arrange for the latter to happen at end of file.
line
Now consider the definition of line:
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
The first alternative is a token which is a newline character; this means
that rpcalc accepts a blank line (and ignores it, since there is no
action). The second alternative is an expression followed by a newline.
This is the alternative that makes rpcalc useful. The semantic value of
the exp grouping is the value of $1 because the exp in
question is the first symbol in the alternative. The action prints this
value, which is the result of the computation the user asked for.
This action is unusual because it does not assign a value to $$. As
a consequence, the semantic value associated with the line is
uninitialized (its value will be unpredictable). This would be a bug if
that value were ever used, but we don't use it: once rpcalc has printed the
value of the user's input line, that value is no longer needed.
expr
The exp grouping has several rules, one for each kind of expression.
The first rule handles the simplest expressions: those that are just numbers.
The second handles an addition-expression, which looks like two expressions
followed by a plus-sign. The third handles subtraction, and so on.
exp: NUM
| exp exp '+' { $$ = $1 + $2; }
| exp exp '-' { $$ = $1 - $2; }
...
;
We have used `|' to join all the rules for exp, but we could
equally well have written them separately:
exp: NUM ;
exp: exp exp '+' { $$ = $1 + $2; } ;
exp: exp exp '-' { $$ = $1 - $2; } ;
...
Most of the rules have actions that compute the value of the expression in
terms of the value of its parts. For example, in the rule for addition,
$1 refers to the first component exp and $2 refers to
the second one. The third component, '+', has no meaningful
associated semantic value, but if it had one you could refer to it as
$3. When yyparse recognizes a sum expression using this
rule, the sum of the two subexpressions' values is produced as the value of
the entire expression. See section Actions.
You don't have to give an action for every rule. When a rule has no
action, Bison by default copies the value of $1 into $$.
This is what happens in the first rule (the one that uses NUM).
The formatting shown here is the recommended convention, but Bison does not require it. You can add or change whitespace as much as you wish. For example, this:
exp : NUM | exp exp '+' {$$ = $1 + $2; } | ...
means the same thing as this:
exp: NUM
| exp exp '+' { $$ = $1 + $2; }
| ...
The latter, however, is much more readable.
rpcalc Lexical Analyzer
The lexical analyzer's job is low-level parsing: converting characters or
sequences of characters into tokens. The Bison parser gets its tokens by
calling the lexical analyzer. See section The Lexical Analyzer Function yylex.
Only a simple lexical analyzer is needed for the RPN calculator. This
lexical analyzer skips blanks and tabs, then reads in numbers as
double and returns them as NUM tokens. Any other character
that isn't part of a number is a separate token. Note that the token-code
for such a single-character token is the character itself.
The return value of the lexical analyzer function is a numeric code which
represents a token type. The same text used in Bison rules to stand for
this token type is also a C expression for the numeric code for the type.
This works in two ways. If the token type is a character literal, then its
numeric code is the ASCII code for that character; you can use the same
character literal in the lexical analyzer to express the number. If the
token type is an identifier, that identifier is defined by Bison as a C
macro whose definition is the appropriate number. In this example,
therefore, NUM becomes a macro for yylex to use.
The semantic value of the token (if it has one) is stored into the global
variable yylval, which is where the Bison parser will look for it.
(The C data type of yylval is YYSTYPE, which was defined
at the beginning of the grammar; see section Declarations for rpcalc.)
A token type code of zero is returned if the end-of-file is encountered. (Bison recognizes any nonpositive value as indicating the end of the input.)
Here is the code for the lexical analyzer:
/* Lexical analyzer returns a double floating point
number on the stack and the token NUM, or the ASCII
character read if not a number. Skips all blanks
and tabs, returns 0 for EOF. */
#include <ctype.h>
yylex ()
{
int c;
/* skip white space */
while ((c = getchar ()) == ' ' || c == '\t')
;
/* process numbers */
if (c == '.' || isdigit (c))
{
ungetc (c, stdin);
scanf ("%lf", &yylval);
return NUM;
}
/* return end-of-file */
if (c == EOF)
return 0;
/* return single chars */
return c;
}
In keeping with the spirit of this example, the controlling function is
kept to the bare minimum. The only requirement is that it call
yyparse to start the process of parsing.
main ()
{
yyparse ();
}
When yyparse detects a syntax error, it calls the error reporting
function yyerror to print an error message (usually but not always
"parse error"). It is up to the programmer to supply yyerror
(see section Parser C-Language Interface), so here is the definition we will use:
#include <stdio.h>
yyerror (s) /* Called by yyparse on error */
char *s;
{
printf ("%s\n", s);
}
After yyerror returns, the Bison parser may recover from the error
and continue parsing if the grammar contains a suitable error rule
(see section Error Recovery). Otherwise, yyparse returns nonzero. We
have not written any error rules in this example, so any invalid input will
cause the calculator program to exit. This is not clean behavior for a
real calculator, but it is adequate in the first example.
Before running Bison to produce a parser, we need to decide how to arrange
all the source code in one or more source files. For such a simple example,
the easiest thing is to put everything in one file. The definitions of
yylex, yyerror and main go at the end, in the
"additional C code" section of the file (see section The Overall Layout of a Bison Grammar).
For a large project, you would probably have several source files, and use
make to arrange to recompile them.
With all the source in a single file, you use the following command to convert it into a parser file:
bison file_name.y
In this example the file was called `rpcalc.y' (for "Reverse Polish
CALCulator"). Bison produces a file named `file_name.tab.c',
removing the `.y' from the original file name. The file output by
Bison contains the source code for yyparse. The additional
functions in the input file (yylex, yyerror and main)
are copied verbatim to the output.
Here is how to compile and run the parser file:
# List files in current directory.
% ls
rpcalc.tab.c rpcalc.y
# Compile the Bison parser.
# `-lm' tells compiler to search math library for pow.
% cc rpcalc.tab.c -lm -o rpcalc
# List files again.
% ls
rpcalc rpcalc.tab.c rpcalc.y
The file `rpcalc' now contains the executable code. Here is an
example session using rpcalc.
% rpcalc 4 9 + 13 3 7 + 3 4 5 *+- -13 3 7 + 3 4 5 * + - n Note the unary minus, `n' 13 5 6 / 4 n + -3.166666667 3 4 ^ Exponentiation 81 ^D End-of-file indicator %
calcWe now modify rpcalc to handle infix operators instead of postfix. Infix notation involves the concept of operator precedence and the need for parentheses nested to arbitrary depth. Here is the Bison code for `calc.y', an infix desk-top calculator.
/* Infix notation calculator--calc */
%{
#define YYSTYPE double
#include <math.h>
%}
/* BISON Declarations */
%token NUM
%left '-' '+'
%left '*' '/'
%left NEG /* negation--unary minus */
%right '^' /* exponentiation */
/* Grammar follows */
%%
input: /* empty string */
| input line
;
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
exp: NUM { $$ = $1; }
| exp '+' exp { $$ = $1 + $3; }
| exp '-' exp { $$ = $1 - $3; }
| exp '*' exp { $$ = $1 * $3; }
| exp '/' exp { $$ = $1 / $3; }
| '-' exp %prec NEG { $$ = -$2; }
| exp '^' exp { $$ = pow ($1, $3); }
| '(' exp ')' { $$ = $2; }
;
%%
The functions yylex, yyerror and main can be the same
as before.
There are two important new features shown in this code.
In the second section (Bison declarations), %left declares token
types and says they are left-associative operators. The declarations
%left and %right (right associativity) take the place of
%token which is used to declare a token type name without
associativity. (These tokens are single-character literals, which
ordinarily don't need to be declared. We declare them here to specify
the associativity.)
Operator precedence is determined by the line ordering of the
declarations; the higher the line number of the declaration (lower on
the page or screen), the higher the precedence. Hence, exponentiation
has the highest precedence, unary minus (NEG) is next, followed
by `*' and `/', and so on. See section Operator Precedence.
The other important new feature is the %prec in the grammar section
for the unary minus operator. The %prec simply instructs Bison that
the rule `| '-' exp' has the same precedence as NEG---in this
case the next-to-highest. See section Context-Dependent Precedence.
Here is a sample run of `calc.y':
% calc 4 + 4.5 - (34/(8*3+-3)) 6.880952381 -56 + 2 -54 3 ^ 2 9
Up to this point, this manual has not addressed the issue of error
recovery---how to continue parsing after the parser detects a syntax
error. All we have handled is error reporting with yyerror. Recall
that by default yyparse returns after calling yyerror. This
means that an erroneous input line causes the calculator program to exit.
Now we show how to rectify this deficiency.
The Bison language itself includes the reserved word error, which
may be included in the grammar rules. In the example below it has
been added to one of the alternatives for line:
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
| error '\n' { yyerrok; }
;
This addition to the grammar allows for simple error recovery in the event
of a parse error. If an expression that cannot be evaluated is read, the
error will be recognized by the third rule for line, and parsing
will continue. (The yyerror function is still called upon to print
its message as well.) The action executes the statement yyerrok, a
macro defined automatically by Bison; its meaning is that error recovery is
complete (see section Error Recovery). Note the difference between
yyerrok and yyerror; neither one is a misprint.
This form of error recovery deals with syntax errors. There are other
kinds of errors; for example, division by zero, which raises an exception
signal that is normally fatal. A real calculator program must handle this
signal and use longjmp to return to main and resume parsing
input lines; it would also have to discard the rest of the current line of
input. We won't discuss this issue further because it is not specific to
Bison programs.
mfcalc
Now that the basics of Bison have been discussed, it is time to move on to
a more advanced problem. The above calculators provided only five
functions, `+', `-', `*', `/' and `^'. It would
be nice to have a calculator that provides other mathematical functions such
as sin, cos, etc.
It is easy to add new operators to the infix calculator as long as they are
only single-character literals. The lexical analyzer yylex passes
back all non-number characters as tokens, so new grammar rules suffice for
adding a new operator. But we want something more flexible: built-in
functions whose syntax has this form:
function_name (argument)
At the same time, we will add memory to the calculator, by allowing you to create named variables, store values in them, and use them later. Here is a sample session with the multi-function calculator:
% mfcalc pi = 3.141592653589 3.1415926536 sin(pi) 0.0000000000 alpha = beta1 = 2.3 2.3000000000 alpha 2.3000000000 ln(alpha) 0.8329091229 exp(ln(beta1)) 2.3000000000 %
Note that multiple assignment and nested function calls are permitted.
mfcalcHere are the C and Bison declarations for the multi-function calculator.
%{
#include <math.h> /* For math functions, cos(), sin(), etc. */
#include "calc.h" /* Contains definition of `symrec' */
%}
%union {
double val; /* For returning numbers. */
symrec *tptr; /* For returning symbol-table pointers */
}
%token <val> NUM /* Simple double precision number */
%token <tptr> VAR FNCT /* Variable and Function */
%type <val> exp
%right '='
%left '-' '+'
%left '*' '/'
%left NEG /* Negation--unary minus */
%right '^' /* Exponentiation */
/* Grammar follows */
%%
The above grammar introduces only two new features of the Bison language. These features allow semantic values to have various data types (see section More Than One Value Type).
The %union declaration specifies the entire list of possible types;
this is instead of defining YYSTYPE. The allowable types are now
double-floats (for exp and NUM) and pointers to entries in
the symbol table. See section The Collection of Value Types.
Since values can now have various types, it is necessary to associate a
type with each grammar symbol whose semantic value is used. These symbols
are NUM, VAR, FNCT, and exp. Their
declarations are augmented with information about their data type (placed
between angle brackets).
The Bison construct %type is used for declaring nonterminal symbols,
just as %token is used for declaring token types. We have not used
%type before because nonterminal symbols are normally declared
implicitly by the rules that define them. But exp must be declared
explicitly so we can specify its value type. See section Nonterminal Symbols.
mfcalc
Here are the grammar rules for the multi-function calculator.
Most of them are copied directly from calc; three rules,
those which mention VAR or FNCT, are new.
input: /* empty */
| input line
;
line:
'\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
| error '\n' { yyerrok; }
;
exp: NUM { $$ = $1; }
| VAR { $$ = $1->value.var; }
| VAR '=' exp { $$ = $3; $1->value.var = $3; }
| FNCT '(' exp ')' { $$ = (*($1->value.fnctptr))($3); }
| exp '+' exp { $$ = $1 + $3; }
| exp '-' exp { $$ = $1 - $3; }
| exp '*' exp { $$ = $1 * $3; }
| exp '/' exp { $$ = $1 / $3; }
| '-' exp %prec NEG { $$ = -$2; }
| exp '^' exp { $$ = pow ($1, $3); }
| '(' exp ')' { $$ = $2; }
;
/* End of grammar */
%%
mfcalc Symbol TableThe multi-function calculator requires a symbol table to keep track of the names and meanings of variables and functions. This doesn't affect the grammar rules (except for the actions) or the Bison declarations, but it requires some additional C functions for support.
The symbol table itself consists of a linked list of records. Its definition, which is kept in the header `calc.h', is as follows. It provides for either functions or variables to be placed in the table.
/* Data type for links in the chain of symbols. */
struct symrec
{
char *name; /* name of symbol */
int type; /* type of symbol: either VAR or FNCT */
union {
double var; /* value of a VAR */
double (*fnctptr)(); /* value of a FNCT */
} value;
struct symrec *next; /* link field */
};
typedef struct symrec symrec;
/* The symbol table: a chain of `struct symrec'. */
extern symrec *sym_table;
symrec *putsym ();
symrec *getsym ();
The new version of main includes a call to init_table, a
function that initializes the symbol table. Here it is, and
init_table as well:
#include <stdio.h>
main ()
{
init_table ();
yyparse ();
}
yyerror (s) /* Called by yyparse on error */
char *s;
{
printf ("%s\n", s);
}
struct init
{
char *fname;
double (*fnct)();
};
struct init arith_fncts[]
= {
"sin", sin,
"cos", cos,
"atan", atan,
"ln", log,
"exp", exp,
"sqrt", sqrt,
0, 0
};
/* The symbol table: a chain of `struct symrec'. */
symrec *sym_table = (symrec *)0;
init_table () /* puts arithmetic functions in table. */
{
int i;
symrec *ptr;
for (i = 0; arith_fncts[i].fname != 0; i++)
{
ptr = putsym (arith_fncts[i].fname, FNCT);
ptr->value.fnctptr = arith_fncts[i].fnct;
}
}
By simply editing the initialization list and adding the necessary include files, you can add additional functions to the calculator.
Two important functions allow look-up and installation of symbols in the
symbol table. The function putsym is passed a name and the type
(VAR or FNCT) of the object to be installed. The object is
linked to the front of the list, and a pointer to the object is returned.
The function getsym is passed the name of the symbol to look up. If
found, a pointer to that symbol is returned; otherwise zero is returned.
symrec *
putsym (sym_name,sym_type)
char *sym_name;
int sym_type;
{
symrec *ptr;
ptr = (symrec *) malloc (sizeof (symrec));
ptr->name = (char *) malloc (strlen (sym_name) + 1);
strcpy (ptr->name,sym_name);
ptr->type = sym_type;
ptr->value.var = 0; /* set value to 0 even if fctn. */
ptr->next = (struct symrec *)sym_table;
sym_table = ptr;
return ptr;
}
symrec *
getsym (sym_name)
char *sym_name;
{
symrec *ptr;
for (ptr = sym_table; ptr != (symrec *) 0;
ptr = (symrec *)ptr->next)
if (strcmp (ptr->name,sym_name) == 0)
return ptr;
return 0;
}
The function yylex must now recognize variables, numeric values, and
the single-character arithmetic operators. Strings of alphanumeric
characters with a leading nondigit are recognized as either variables or
functions depending on what the symbol table says about them.
The string is passed to getsym for look up in the symbol table. If
the name appears in the table, a pointer to its location and its type
(VAR or FNCT) is returned to yyparse. If it is not
already in the table, then it is installed as a VAR using
putsym. Again, a pointer and its type (which must be VAR) is
returned to yyparse.
No change is needed in the handling of numeric values and arithmetic
operators in yylex.
#include <ctype.h>
yylex ()
{
int c;
/* Ignore whitespace, get first nonwhite character. */
while ((c = getchar ()) == ' ' || c == '\t');
if (c == EOF)
return 0;
/* Char starts a number => parse the number. */
if (c == '.' || isdigit (c))
{
ungetc (c, stdin);
scanf ("%lf", &yylval.val);
return NUM;
}
/* Char starts an identifier => read the name. */
if (isalpha (c))
{
symrec *s;
static char *symbuf = 0;
static int length = 0;
int i;
/* Initially make the buffer long enough
for a 40-character symbol name. */
if (length == 0)
length = 40, symbuf = (char *)malloc (length + 1);
i = 0;
do
{
/* If buffer is full, make it bigger. */
if (i == length)
{
length *= 2;
symbuf = (char *)realloc (symbuf, length + 1);
}
/* Add this character to the buffer. */
symbuf[i++] = c;
/* Get another character. */
c = getchar ();
}
while (c != EOF && isalnum (c));
ungetc (c, stdin);
symbuf[i] = '\0';
s = getsym (symbuf);
if (s == 0)
s = putsym (symbuf, VAR);
yylval.tptr = s;
return s->type;
}
/* Any other character is a token by itself. */
return c;
}
This program is both powerful and flexible. You may easily add new
functions, and it is a simple job to modify this code to install predefined
variables such as pi or e as well.
init_table to add these constants to the symbol table.
It will be easiest to give the constants type VAR.
Bison takes as input a context-free grammar specification and produces a C-language function that recognizes correct instances of the grammar.
The Bison grammar input file conventionally has a name ending in `.y'.
A Bison grammar file has four main sections, shown here with the appropriate delimiters:
%{
C declarations
%}
Bison declarations
%%
Grammar rules
%%
Additional C code
Comments enclosed in `/* ... */' may appear in any of the sections.
The C declarations section contains macro definitions and
declarations of functions and variables that are used in the actions in the
grammar rules. These are copied to the beginning of the parser file so
that they precede the definition of yyparse. You can use
`#include' to get the declarations from a header file. If you don't
need any C declarations, you may omit the `%{' and `%}'
delimiters that bracket this section.
The Bison declarations section contains declarations that define terminal and nonterminal symbols, specify precedence, and so on. In some simple grammars you may not need any declarations. See section Bison Declarations.
The grammar rules section contains one or more Bison grammar rules, and nothing else. See section Syntax of Grammar Rules.
There must always be at least one grammar rule, and the first `%%' (which precedes the grammar rules) may never be omitted even if it is the first thing in the file.
The additional C code section is copied verbatim to the end of
the parser file, just as the C declarations section is copied to
the beginning. This is the most convenient place to put anything
that you want to have in the parser file but which need not come before
the definition of yyparse. For example, the definitions of
yylex and yyerror often go here. See section Parser C-Language Interface.
If the last section is empty, you may omit the `%%' that separates it from the grammar rules.
The Bison parser itself contains many static variables whose names start with `yy' and many macros whose names start with `YY'. It is a good idea to avoid using any such names (except those documented in this manual) in the additional C code section of the grammar file.
Symbols in Bison grammars represent the grammatical classifications of the language.
A terminal symbol (also known as a token type) represents a
class of syntactically equivalent tokens. You use the symbol in grammar
rules to mean that a token in that class is allowed. The symbol is
represented in the Bison parser by a numeric code, and the yylex
function returns a token type code to indicate what kind of token has been
read. You don't need to know what the code value is; you can use the
symbol to stand for it.
A nonterminal symbol stands for a class of syntactically equivalent groupings. The symbol name is used in writing grammar rules. By convention, it should be all lower case.
Symbol names can contain letters, digits (not at the beginning), underscores and periods. Periods make sense only in nonterminals.
There are three ways of writing terminal symbols in the grammar:
%token. See section Token Type Names.
'+' is a character token type. A
character token type doesn't need to be declared unless you need to
specify its semantic value data type (see section Data Types of Semantic Values), associativity, or precedence (see section Operator Precedence).
By convention, a character token type is used only to represent a
token that consists of that particular character. Thus, the token
type '+' is used to represent the character `+' as a
token. Nothing enforces this convention, but if you depart from it,
your program will confuse other readers.
All the usual escape sequences used in character literals in C can be
used in Bison as well, but you must not use the null character as a
character literal because its ASCII code, zero, is the code yylex
returns for end-of-input (see section Calling Convention for yylex).
"<=" is a literal string token. A literal string token
doesn't need to be declared unless you need to specify its semantic
value data type (see section Data Types of Semantic Values), associativity, precedence
(see section Operator Precedence).
You can associate the literal string token with a symbolic name as an
alias, using the %token declaration (see section Token Type Names). If you don't do that, the lexical analyzer has to
retrieve the token number for the literal string token from the
yytname table (see section Calling Convention for yylex).
WARNING: literal string tokens do not work in Yacc.
By convention, a literal string token is used only to represent a token
that consists of that particular string. Thus, you should use the token
type "<=" to represent the string `<=' as a token. Bison
does not enforces this convention, but if you depart from it, people who
read your program will be confused.
All the escape sequences used in string literals in C can be used in
Bison as well. A literal string token must contain two or more
characters; for a token containing just one character, use a character
token (see above).
How you choose to write a terminal symbol has no effect on its grammatical meaning. That depends only on where it appears in rules and on when the parser function returns that symbol.
The value returned by yylex is always one of the terminal symbols
(or 0 for end-of-input). Whichever way you write the token type in the
grammar rules, you write it the same way in the definition of yylex.
The numeric code for a character token type is simply the ASCII code for
the character, so yylex can use the identical character constant to
generate the requisite code. Each named token type becomes a C macro in
the parser file, so yylex can use the name to stand for the code.
(This is why periods don't make sense in terminal symbols.)
See section Calling Convention for yylex.
If yylex is defined in a separate file, you need to arrange for the
token-type macro definitions to be available there. Use the `-d'
option when you run Bison, so that it will write these macro definitions
into a separate header file `name.tab.h' which you can include
in the other source files that need it. See section Invoking Bison.
The symbol error is a terminal symbol reserved for error recovery
(see section Error Recovery); you shouldn't use it for any other purpose.
In particular, yylex should never return this value.
A Bison grammar rule has the following general form:
result: components...
;
where result is the nonterminal symbol that this rule describes and components are various terminal and nonterminal symbols that are put together by this rule (see section Symbols, Terminal and Nonterminal).
For example,
exp: exp '+' exp
;
says that two groupings of type exp, with a `+' token in between,
can be combined into a larger grouping of type exp.
Whitespace in rules is significant only to separate symbols. You can add extra whitespace as you wish.
Scattered among the components can be actions that determine the semantics of the rule. An action looks like this:
{C statements}
Usually there is only one action and it follows the components. See section Actions.
Multiple rules for the same result can be written separately or can be joined with the vertical-bar character `|' as follows:
result: rule1-components...
| rule2-components...
...
;
They are still considered distinct rules even when joined in this way.
If components in a rule is empty, it means that result can
match the empty string. For example, here is how to define a
comma-separated sequence of zero or more exp groupings:
expseq: /* empty */
| expseq1
;
expseq1: exp
| expseq1 ',' exp
;
It is customary to write a comment `/* empty */' in each rule with no components.
A rule is called recursive when its result nonterminal appears also on its right hand side. Nearly all Bison grammars need to use recursion, because that is the only way to define a sequence of any number of somethings. Consider this recursive definition of a comma-separated sequence of one or more expressions:
expseq1: exp
| expseq1 ',' exp
;
Since the recursive use of expseq1 is the leftmost symbol in the
right hand side, we call this left recursion. By contrast, here
the same construct is defined using right recursion:
expseq1: exp
| exp ',' expseq1
;
Any kind of sequence can be defined using either left recursion or right recursion, but you should always use left recursion, because it can parse a sequence of any number of elements with bounded stack space. Right recursion uses up space on the Bison stack in proportion to the number of elements in the sequence, because all the elements must be shifted onto the stack before the rule can be applied even once. See section The Bison Parser Algorithm, for further explanation of this.
Indirect or mutual recursion occurs when the result of the rule does not appear directly on its right hand side, but does appear in rules for other nonterminals which do appear on its right hand side.
For example:
expr: primary
| primary '+' primary
;
primary: constant
| '(' expr ')'
;
defines two mutually-recursive nonterminals, since each refers to the other.
The grammar rules for a language determine only the syntax. The semantics are determined by the semantic values associated with various tokens and groupings, and by the actions taken when various groupings are recognized.
For example, the calculator calculates properly because the value associated with each expression is the proper number; it adds properly because the action for the grouping `x + y' is to add the numbers associated with x and y.
In a simple program it may be sufficient to use the same data type for the semantic values of all language constructs. This was true in the RPN and infix calculator examples (see section Reverse Polish Notation Calculator).
Bison's default is to use type int for all semantic values. To
specify some other type, define YYSTYPE as a macro, like this:
#define YYSTYPE double
This macro definition must go in the C declarations section of the grammar file (see section Outline of a Bison Grammar).
In most programs, you will need different data types for different kinds
of tokens and groupings. For example, a numeric constant may need type
int or long, while a string constant needs type char *,
and an identifier might need a pointer to an entry in the symbol table.
To use more than one data type for semantic values in one parser, Bison requires you to do two things:
%union Bison declaration (see section The Collection of Value Types).
%token Bison declaration (see section Token Type Names) and for groupings
with the %type Bison declaration (see section Nonterminal Symbols).
An action accompanies a syntactic rule and contains C code to be executed each time an instance of that rule is recognized. The task of most actions is to compute a semantic value for the grouping built by the rule from the semantic values associated with tokens or smaller groupings.
An action consists of C statements surrounded by braces, much like a compound statement in C. It can be placed at any position in the rule; it is executed at that position. Most rules have just one action at the end of the rule, following all the components. Actions in the middle of a rule are tricky and used only for special purposes (see section Actions in Mid-Rule).
The C code in an action can refer to the semantic values of the components
matched by the rule with the construct $n, which stands for
the value of the nth component. The semantic value for the grouping
being constructed is $$. (Bison translates both of these constructs
into array element references when it copies the actions into the parser
file.)
Here is a typical example:
exp: ...
| exp '+' exp
{ $$ = $1 + $3; }
This rule constructs an exp from two smaller exp groupings
connected by a plus-sign token. In the action, $1 and $3
refer to the semantic values of the two component exp groupings,
which are the first and third symbols on the right hand side of the rule.
The sum is stored into $$ so that it becomes the semantic value of
the addition-expression just recognized by the rule. If there were a
useful semantic value associated with the `+' token, it could be
referred to as $2.
If you don't specify an action for a rule, Bison supplies a default:
$$ = $1. Thus, the value of the first symbol in the rule becomes
the value of the whole rule. Of course, the default rule is valid only
if the two data types match. There is no meaningful default action for
an empty rule; every empty rule must have an explicit action unless the
rule's value does not matter.
$n with n zero or negative is allowed for reference
to tokens and groupings on the stack before those that match the
current rule. This is a very risky practice, and to use it reliably
you must be certain of the context in which the rule is applied. Here
is a case in which you can use this reliably:
foo: expr bar '+' expr { ... }
| expr bar '-' expr { ... }
;
bar: /* empty */
{ previous_expr = $0; }
;
As long as bar is used only in the fashion shown here, $0
always refers to the expr which precedes bar in the
definition of foo.
If you have chosen a single data type for semantic values, the $$
and $n constructs always have that data type.
If you have used %union to specify a variety of data types, then you
must declare a choice among these types for each terminal or nonterminal
symbol that can have a semantic value. Then each time you use $$ or
$n, its data type is determined by which symbol it refers to
in the rule. In this example,
exp: ...
| exp '+' exp
{ $$ = $1 + $3; }
$1 and $3 refer to instances of exp, so they all
have the data type declared for the nonterminal symbol exp. If
$2 were used, it would have the data type declared for the
terminal symbol '+', whatever that might be.
Alternatively, you can specify the data type when you refer to the value, by inserting `<type>' after the `$' at the beginning of the reference. For example, if you have defined types as shown here:
%union {
int itype;
double dtype;
}
then you can write $<itype>1 to refer to the first subunit of the
rule as an integer, or $<dtype>1 to refer to it as a double.
Occasionally it is useful to put an action in the middle of a rule. These actions are written just like usual end-of-rule actions, but they are executed before the parser even recognizes the following components.
A mid-rule action may refer to the components preceding it using
$n, but it may not refer to subsequent components because
it is run before they are parsed.
The mid-rule action itself counts as one of the components of the rule.
This makes a difference when there is another action later in the same rule
(and usually there is another at the end): you have to count the actions
along with the symbols when working out which number n to use in
$n.
The mid-rule action can also have a semantic value. The action can set
its value with an assignment to $$, and actions later in the rule
can refer to the value using $n. Since there is no symbol
to name the action, there is no way to declare a data type for the value
in advance, so you must use the `$<...>' construct to specify a
data type each time you refer to this value.
There is no way to set the value of the entire rule with a mid-rule
action, because assignments to $$ do not have that effect. The
only way to set the value for the entire rule is with an ordinary action
at the end of the rule.
Here is an example from a hypothetical compiler, handling a let
statement that looks like `let (variable) statement' and
serves to create a variable named variable temporarily for the
duration of statement. To parse this construct, we must put
variable into the symbol table while statement is parsed, then
remove it afterward. Here is how it is done:
stmt: LET '(' var ')'
{ $<context>$ = push_context ();
declare_variable ($3); }
stmt { $$ = $6;
pop_context ($<context>5); }
As soon as `let (variable)' has been recognized, the first
action is run. It saves a copy of the current semantic context (the
list of accessible variables) as its semantic value, using alternative
context in the data-type union. Then it calls
declare_variable to add the new variable to that list. Once the
first action is finished, the embedded statement stmt can be
parsed. Note that the mid-rule action is component number 5, so the
`stmt' is component number 6.
After the embedded statement is parsed, its semantic value becomes the
value of the entire let-statement. Then the semantic value from the
earlier action is used to restore the prior list of variables. This
removes the temporary let-variable from the list so that it won't
appear to exist while the rest of the program is parsed.
Taking action before a rule is completely recognized often leads to conflicts since the parser must commit to a parse in order to execute the action. For example, the following two rules, without mid-rule actions, can coexist in a working parser because the parser can shift the open-brace token and look at what follows before deciding whether there is a declaration or not:
compound: '{' declarations statements '}'
| '{' statements '}'
;
But when we add a mid-rule action as follows, the rules become nonfunctional:
compound: { prepare_for_local_variables (); }
'{' declarations statements '}'
| '{' statements '}'
;
Now the parser is forced to decide whether to run the mid-rule action when it has read no farther than the open-brace. In other words, it must commit to using one rule or the other, without sufficient information to do it correctly. (The open-brace token is what is called the look-ahead token at this time, since the parser is still deciding what to do about it. See section Look-Ahead Tokens.)
You might think that you could correct the problem by putting identical actions into the two rules, like this:
compound: { prepare_for_local_variables (); }
'{' declarations statements '}'
| { prepare_for_local_variables (); }
'{' statements '}'
;
But this does not help, because Bison does not realize that the two actions are identical. (Bison never tries to understand the C code in an action.)
If the grammar is such that a declaration can be distinguished from a statement by the first token (which is true in C), then one solution which does work is to put the action after the open-brace, like this:
compound: '{' { prepare_for_local_variables (); }
declarations statements '}'
| '{' statements '}'
;
Now the first token of the following declaration or statement, which would in any case tell Bison which rule to use, can still do so.
Another solution is to bury the action inside a nonterminal symbol which serves as a subroutine:
subroutine: /* empty */
{ prepare_for_local_variables (); }
;
compound: subroutine
'{' declarations statements '}'
| subroutine
'{' statements '}'
;
Now Bison can execute the action in the rule for subroutine without
deciding which rule for compound it will eventually use. Note that
the action is now at the end of its rule. Any mid-rule action can be
converted to an end-of-rule action in this way, and this is what Bison
actually does to implement mid-rule actions.
The Bison declarations section of a Bison grammar defines the symbols used in formulating the grammar and the data types of semantic values. See section Symbols, Terminal and Nonterminal.
All token type names (but not single-character literal tokens such as
'+' and '*') must be declared. Nonterminal symbols must be
declared if you need to specify which data type to use for the semantic
value (see section More Than One Value Type).
The first rule in the file also specifies the start symbol, by default. If you want some other symbol to be the start symbol, you must declare it explicitly (see section Languages and Context-Free Grammars).
The basic way to declare a token type name (terminal symbol) is as follows:
%token name
Bison will convert this into a #define directive in
the parser, so that the function yylex (if it is in this file)
can use the name name to stand for this token type's code.
Alternatively, you can use %left, %right, or %nonassoc
instead of %token, if you wish to specify precedence.
See section Operator Precedence.
You can explicitly specify the numeric code for a token type by appending an integer value in the field immediately following the token name:
%token NUM 300
It is generally best, however, to let Bison choose the numeric codes for all token types. Bison will automatically select codes that don't conflict with each other or with ASCII characters.
In the event that the stack type is a union, you must augment the
%token or other token declaration to include the data type
alternative delimited by angle-brackets (see section More Than One Value Type).
For example:
%union { /* define stack type */
double val;
symrec *tptr;
}
%token <val> NUM /* define token NUM and its type */
You can associate a literal string token with a token type name by
writing the literal string at the end of a %token
declaration which declares the name. For example:
%token arrow "=>"
For example, a grammar for the C language might specify these names with equivalent literal string tokens:
%token <operator> OR "||" %token <operator> LE 134 "<=" %left OR "<="
Once you equate the literal string and the token name, you can use them
interchangeably in further declarations or the grammar rules. The
yylex function can use the token name or the literal string to
obtain the token type code number (see section Calling Convention for yylex).
Use the %left, %right or %nonassoc declaration to
declare a token and specify its precedence and associativity, all at
once. These are called precedence declarations.
See section Operator Precedence, for general information on operator precedence.
The syntax of a precedence declaration is the same as that of
%token: either
%left symbols...
or
%left <type> symbols...
And indeed any of these declarations serves the purposes of %token.
But in addition, they specify the associativity and relative precedence for
all the symbols:
%left specifies
left-associativity (grouping x with y first) and
%right specifies right-associativity (grouping y with
z first). %nonassoc specifies no associativity, which
means that `x op y op z' is
considered a syntax error.
The %union declaration specifies the entire collection of possible
data types for semantic values. The keyword %union is followed by a
pair of braces containing the same thing that goes inside a union in
C.
For example:
%union {
double val;
symrec *tptr;
}
This says that the two alternative types are double and symrec
*. They are given names val and tptr; these names are used
in the %token and %type declarations to pick one of the types
for a terminal or nonterminal symbol (see section Nonterminal Symbols).
Note that, unlike making a union declaration in C, you do not write
a semicolon after the closing brace.
When you use %union to specify multiple value types, you must
declare the value type of each nonterminal symbol for which values are
used. This is done with a %type declaration, like this:
%type <type> nonterminal...
Here nonterminal is the name of a nonterminal symbol, and type
is the name given in the %union to the alternative that you want
(see section The Collection of Value Types). You can give any number of nonterminal symbols in
the same %type declaration, if they have the same value type. Use
spaces to separate the symbol names.
You can also declare the value type of a terminal symbol. To do this,
use the same <type> construction in a declaration for the
terminal symbol. All kinds of token declarations allow
<type>.
Bison normally warns if there are any conflicts in the grammar
(see section Shift/Reduce Conflicts), but most real grammars have harmless shift/reduce
conflicts which are resolved in a predictable way and would be difficult to
eliminate. It is desirable to suppress the warning about these conflicts
unless the number of conflicts changes. You can do this with the
%expect declaration.
The declaration looks like this:
%expect n
Here n is a decimal integer. The declaration says there should be no warning if there are n shift/reduce conflicts and no reduce/reduce conflicts. The usual warning is given if there are either more or fewer conflicts, or if there are any reduce/reduce conflicts.
In general, using %expect involves these steps:
%expect. Use the `-v' option
to get a verbose list of where the conflicts occur. Bison will also
print the number of conflicts.
%expect declaration, copying the number n from the
number which Bison printed.
Now Bison will stop annoying you about the conflicts you have checked, but it will warn you again if changes in the grammar result in additional conflicts.
Bison assumes by default that the start symbol for the grammar is the first
nonterminal specified in the grammar specification section. The programmer
may override this restriction with the %start declaration as follows:
%start symbol
A reentrant program is one which does not alter in the course of execution; in other words, it consists entirely of pure (read-only) code. Reentrancy is important whenever asynchronous execution is possible; for example, a nonreentrant program may not be safe to call from a signal handler. In systems with multiple threads of control, a nonreentrant program must be called only within interlocks.
The Bison parser is not normally a reentrant program, because it uses
statically allocated variables for communication with yylex. These
variables include yylval and yylloc.
The Bison declaration %pure_parser says that you want the parser
to be reentrant. It looks like this:
%pure_parser
The effect is that the two communication variables become local
variables in yyparse, and a different calling convention is used
for the lexical analyzer function yylex. See section Calling Conventions for Pure Parsers, for the details of this. The
variable yynerrs also becomes local in yyparse
(see section The Error Reporting Function yyerror).
The convention for calling yyparse itself is unchanged.
Here is a summary of all Bison declarations:
%union
%token
%right
%left
%nonassoc
%type
%start
%expect
%pure_parser
%no_lines
#line preprocessor commands in the parser
file. Ordinarily Bison writes these commands in the parser file so that
the C compiler and debuggers will associate errors and object code with
your source file (the grammar file). This directive causes them to
associate errors with the parser file, treating it an independent source
file in its own right.
%raw
%token_table
yytname; yytname[i] is the name of the
token whose internal Bison token code number is i. The first three
elements of yytname are always "$", "error", and
"$illegal"; after these come the symbols defined in the grammar
file.
For single-character literal tokens and literal string tokens, the name
in the table includes the single-quote or double-quote characters: for
example, "'+'" is a single-character literal and "\"<=\""
is a literal string token. All the characters of the literal string
token appear verbatim in the string found in the table; even
double-quote characters are not escaped. For example, if the token
consists of three characters `*"*', its string in yytname
contains `"*"*"'. (In C, that would be written as
"\"*\"*\"").
When you specify %token_table, Bison also generates macro
definitions for macros YYNTOKENS, YYNNTS, and
YYNRULES, and YYNSTATES:
YYNTOKENS
YYNNTS
YYNRULES
YYNSTATES
Most programs that use Bison parse only one language and therefore contain
only one Bison parser. But what if you want to parse more than one
language with the same program? Then you need to avoid a name conflict
between different definitions of yyparse, yylval, and so on.
The easy way to do this is to use the option `-p prefix' (see section Invoking Bison). This renames the interface functions and variables of the Bison parser to start with prefix instead of `yy'. You can use this to give each parser distinct names that do not conflict.
The precise list of symbols renamed is yyparse, yylex,
yyerror, yynerrs, yylval, yychar and
yydebug. For example, if you use `-p c', the names become
cparse, clex, and so on.
All the other variables and macros associated with Bison are not
renamed. These others are not global; there is no conflict if the same
name is used in different parsers. For example, YYSTYPE is not
renamed, but defining this in different ways in different parsers causes
no trouble (see section Data Types of Semantic Values).
The `-p' option works by adding macro definitions to the beginning
of the parser source file, defining yyparse as
prefixparse, and so on. This effectively substitutes one
name for the other in the entire parser file.
The Bison parser is actually a C function named yyparse. Here we
describe the interface conventions of yyparse and the other
functions that it needs to use.
Keep in mind that the parser uses many C identifiers starting with `yy' and `YY' for internal purposes. If you use such an identifier (aside from those in this manual) in an action or in additional C code in the grammar file, you are likely to run into trouble.
yyparse
You call the function yyparse to cause parsing to occur. This
function reads tokens, executes actions, and ultimately returns when it
encounters end-of-input or an unrecoverable syntax error. You can also
write an action which directs yyparse to return immediately without
reading further.
The value returned by yyparse is 0 if parsing was successful (return
is due to end-of-input).
The value is 1 if parsing failed (return is due to a syntax error).
In an action, you can cause immediate return from yyparse by using
these macros:
YYACCEPT
YYABORT
yylex
The lexical analyzer function, yylex, recognizes tokens from
the input stream and returns them to the parser. Bison does not create
this function automatically; you must write it so that yyparse can
call it. The function is sometimes referred to as a lexical scanner.
In simple programs, yylex is often defined at the end of the Bison
grammar file. If yylex is defined in a separate source file, you
need to arrange for the token-type macro definitions to be available there.
To do this, use the `-d' option when you run Bison, so that it will
write these macro definitions into a separate header file
`name.tab.h' which you can include in the other source files
that need it. See section Invoking Bison.
yylex
The value that yylex returns must be the numeric code for the type
of token it has just found, or 0 for end-of-input.
When a token is referred to in the grammar rules by a name, that name
in the parser file becomes a C macro whose definition is the proper
numeric code for that token type. So yylex can use the name
to indicate that type. See section Symbols, Terminal and Nonterminal.
When a token is referred to in the grammar rules by a character literal,
the numeric code for that character is also the code for the token type.
So yylex can simply return that character code. The null character
must not be used this way, because its code is zero and that is what
signifies end-of-input.
Here is an example showing these things:
yylex ()
{
...
if (c == EOF) /* Detect end of file. */
return 0;
...
if (c == '+' || c == '-')
return c; /* Assume token type for `+' is '+'. */
...
return INT; /* Return the type of the token. */
...
}
This interface has been designed so that the output from the lex
utility can be used without change as the definition of yylex.
If the grammar uses literal string tokens, there are two ways that
yylex can determine the token type codes for them:
yylex can use these symbolic names like
all others. In this case, the use of the literal string tokens in
the grammar file has no effect on yylex.
yylex can find the multi-character token in the yytname
table. The index of the token in the table is the token type's code.
The name of a multi-character token is recorded in yytname with a
double-quote, the token's characters, and another double-quote. The
token's characters are not escaped in any way; they appear verbatim in
the contents of the string in the table.
Here's code for looking up a token in yytname, assuming that the
characters of the token are stored in token_buffer.
for (i = 0; i < YYNTOKENS; i++)
{
if (yytname[i] != 0
&& yytname[i][0] == '"'
&& strncmp (yytname[i] + 1, token_buffer, strlen (token_buffer))
&& yytname[i][strlen (token_buffer) + 1] == '"'
&& yytname[i][strlen (token_buffer) + 2] == 0)
break;
}
The yytname table is generated only if you use the
%token_table declaration. See section Bison Declaration Summary.
In an ordinary (nonreentrant) parser, the semantic value of the token must
be stored into the global variable yylval. When you are using
just one data type for semantic values, yylval has that type.
Thus, if the type is int (the default), you might write this in
yylex:
... yylval = value; /* Put value onto Bison stack. */ return INT; /* Return the type of the token. */ ...
When you are using multiple data types, yylval's type is a union
made from the %union declaration (see section The Collection of Value Types). So when
you store a token's value, you must use the proper member of the union.
If the %union declaration looks like this:
%union {
int intval;
double val;
symrec *tptr;
}
then the code in yylex might look like this:
... yylval.intval = value; /* Put value onto Bison stack. */ return INT; /* Return the type of the token. */ ...
If you are using the `@n'-feature (see section Special Features for Use in Actions) in
actions to keep track of the textual locations of tokens and groupings,
then you must provide this information in yylex. The function
yyparse expects to find the textual location of a token just parsed
in the global variable yylloc. So yylex must store the
proper data in that variable. The value of yylloc is a structure
and you need only initialize the members that are going to be used by the
actions. The four members are called first_line,
first_column, last_line and last_column. Note that
the use of this feature makes the parser noticeably slower.
The data type of yylloc has the name YYLTYPE.
When you use the Bison declaration %pure_parser to request a
pure, reentrant parser, the global communication variables yylval
and yylloc cannot be used. (See section A Pure (Reentrant) Parser.) In such parsers the two global variables are replaced by
pointers passed as arguments to yylex. You must declare them as
shown here, and pass the information back by storing it through those
pointers.
yylex (lvalp, llocp)
YYSTYPE *lvalp;
YYLTYPE *llocp;
{
...
*lvalp = value; /* Put value onto Bison stack. */
return INT; /* Return the type of the token. */
...
}
If the grammar file does not use the `@' constructs to refer to
textual positions, then the type YYLTYPE will not be defined. In
this case, omit the second argument; yylex will be called with
only one argument.
If you use a reentrant parser, you can optionally pass additional
parameter information to it in a reentrant way. To do so, define the
macro YYPARSE_PARAM as a variable name. This modifies the
yyparse function to accept one argument, of type void *,
with that name.
When you call yyparse, pass the address of an object, casting the
address to void *. The grammar actions can refer to the contents
of the object by casting the pointer value back to its proper type and
then dereferencing it. Here's an example. Write this in the parser:
%{
struct parser_control
{
int nastiness;
int randomness;
};
#define YYPARSE_PARAM parm
%}
Then call the parser like this:
struct parser_control
{
int nastiness;
int randomness;
};
...
{
struct parser_control foo;
... /* Store proper data in foo. */
value = yyparse ((void *) &foo);
...
}
In the grammar actions, use expressions like this to refer to the data:
((struct parser_control *) parm)->randomness
If you wish to pass the additional parameter data to yylex,
define the macro YYLEX_PARAM just like YYPARSE_PARAM, as
shown here:
%{
struct parser_control
{
int nastiness;
int randomness;
};
#define YYPARSE_PARAM parm
#define YYLEX_PARAM parm
%}
You should then define yylex to accept one additional
argument--the value of parm. (This makes e